Over the years we have been focused on developing content and adding new types of media, to the exclusion of any efforts to raise funds.
As you may be aware, the American Business History Center is a labor of love and we have devoted enormous hours to it without much compensation.
We have not actively solicited donations in years, and the donations we have received have gone in large part to expanding our research library and maintaining the website and newsletter.
We’ve been going for 8 years and have published about 300 stories, videos, charts, and audio files.
Our newsletter garners an estimated 5,000 weekly readers, and the website receives hundreds of thousands of annual visits. All of those metrics continue to grow nicely.
In order to continue to develop content and find new ways to create that content, we could use your financial support.
Please consider making a donation to us by clicking this link. Any amount would be much appreciated.
Thanks,
Gary Hoover
Executive Director
American Business History Center
The United States has been building great infrastructure systems for centuries. Railroads, telegraph lines, telephone networks, cable television, and cellular communications required enormous investment before anyone could be certain how profitable or socially useful they would ultimately become.
Today, similar questions are being asked about the vast sums flowing into data centers for artificial intelligence. Will all that investment pay off? Is AI consuming an unreasonable portion of America’s capital?
A reasonable estimate is that the United States will invest about $2.1 trillion in AI-related data centers and supporting infrastructure through 2030.
Compare what we invested in the railroads. America’s railroad adventure began around 1830 and reached peak route mileage in 1916. Over those 86 years, the nation built 254,037 miles of intercity railroad. Investment in railroad road and equipment had reached approximately $21.1 billion by 1916.
The best available estimates of our nation’s past wealth indicate that national wealth was approximately $2.7 billion when the railroad era began, about $34 billion at its chronological midpoint in 1873, and approximately $220 billion when mileage peaked in 1916.
Adjusting for inflation and the different values of the dollar at those times, the railroad system represented approximately 7.8 times the nation’s wealth at the outset, 69% at the chronological midpoint, and 9.6% at peak mileage.
Going back that far in time means that those numbers are rough estimates. Nevertheless, the order of magnitude is clear: the railroad commitment was immense relative to the resources of America and its people.
Relative to that wealth, our investment in AI data centers will be substantially less. U.S. national wealth was approximately $144.5 trillion in 2023, when the present AI boom became unmistakable. Assuming national wealth grows by about 5% annually, it will reach $185 trillion in 2027 and $214 trillion in 2030.
Adjusting to comparable dollars, the estimated $2.1 trillion AI commitment represents approximately 1.4% of national wealth at the 2023 outset and 1.1% in 2030.
Based on those numbers, AI-related data centers will cost about one-ninth as much as the railroad system relative to national wealth.
We have also invested substantial sums in the original wired telephone network and more recently in the cellular system.
The fixed-line telephone network began in 1876 and reached its maximum number of access lines around 2000, a span of 124 years. Its accumulated plant represented approximately $340 billion in 2000 dollars.
The eventual telephone system equaled roughly 56% of national wealth at the outset, 7.2% at the 1938 midpoint, and 0.7% in 2000. By then, 97.6% of American homes had telephone service.
Commercial cellular service began in 1983. By 2024—41 years later—wireless carriers had invested $734 billion in actual nominal capital expenditures. By 2024 there were 579 million active wireless connections, or 1.7 connections for every American, while 98% of adults owned a cellphone. The cellular investment equaled approximately 2.3% of national wealth at the outset, 1.1% at the midpoint, and 0.7% in 2024.
AI is growing more rapidly than any of these earlier systems. In 2023, 18% of American adults had used ChatGPT. By early 2026, 49% had used an AI chatbot and 24% used one daily. The technology has reached nearly half of American adults only three years into the boom.
But America is not wagering anything remotely like the share of its wealth that it committed to the railroads. The railroad system took 86 years and ultimately represented 9.6% of national wealth at peak mileage. The projected first great phase of AI infrastructure will take approximately seven years from the beginning of the present boom and represent about 1.1% of projected national wealth in 2030. While the total capital going into AI is perhaps three times the amount that will be spent on data centers, is it still no greater than what we spent to build our excellent railroad system.
